Dynamic Interactions Options Store

The Options Store is generated from data or manually created and updated. Follow the process in parameters to configure the options store. This store will be updated during the background process cycle. The options are generated from a feature store or data set that contains the appropriate date for your use-case.

Options Store Example

Below is an example document from an Options Store collection:

{
    "contextual_variable_two": "",
    "beta_zero": 1,
    "date_updated": "2024-06-11T08:30:39.000450Z",
    "epsilon_nominated": 1,
    "date_created": "2024-06-11T08:30:39.000450Z",
    "contextual_variable_one": "",
    "uuid": "2a4c537a-6e76-4383-aa4d-d7644e62fcfe",
    "weighting": 1,
    "alpha_zero": 1,
    "propensity": 0.218988,
    "optionKey": "Enthusiastic",
    "date_time": "2024-08-09T01:23:19.000862Z",
    "alpha": 0.218988,
    "id": "spend_personality_dynamic",
    "beta": 0.781012,
    "arm_reward": 0.00218988,
    "option": "Recommended offer is Enthusiastic",
    "customer": "none"
}

Updating the Options Store

If you want to add new Options to your configuration you will need to refresh your Options Store to include these items. This can be done using either the Workbench or the python package. To add these new options they need to be added to your set up Feature Store. There are two approaches you can follow to perform the refresh:

The entire Options Store can be regenerated. This will reset all of the existing Options to the default value, in this process any new options in the Feature Store will be added to the Options Store. This will remove Options no longer present in the set up Feature store from the Options Store
The existing Options Store can be updated. This will only add new options in the Feature Store to the existing Options Store and will not reset the existing options. This will not remove Options no longer present in the set up Feature Store from the Options Store.

🔥

Note

If the Options Store is regenerated for a running use case the Options Store will continue to be updated by the runtime using the existing logs at the scheduled interval. This means that your regenerated Options Store could largely return to it’s prior state once the runtime updates the Options Store.

Workbench

The Options Store is generated in the Variables tab of the Dynamic Interaction configuration.

Use the Generate button to generate a new options store from the settings. Defaults are extracted from defined Feature Store and the Options Store will be generated. Use the Update button if you have an existing options store that needs updating. It will not re-generate the options store, but only add or update the options that are out of date. All scores will be retained and defaults will be used for added options only.

Python

The Options Store can also be Updated or Generated using the python package. The following example shows how to update the Options Store using the python package. Here we include we give an example of updating the set up Feature Store from an Offer Matrix and then updating the Options Store.

# Import packages
from prediction.apis import data_management_engine as dme
from prediction.apis import online_learning_management as ol
from prediction import jwt_access
import getpass
 
# Connect to the ecosystem.Ai server
ecosystem_password = getpass.getpass("Enter your ecosystem password")
auth = jwt_access.Authenticate("http://ecosystem-server:3001/api", "user@ecosystem.ai", ecosystem_password)
 
# Configure the name of the Dynamic Interaction configuration. The uuid can also be used
deployment_id = "dynamic-recommender"
 
# Update your online learning set up feature store
ol.online_learning_ecosystem_rewards_setup_feature_store(
    auth,
    offer_db = "recommender_demos",
    offer_collection = "offer_matrix",
    offer_name_column = "offer",
    contextual_variables = {"segment_one":["lt-15","gt-15"],"segment_two":["lt-50","50-250","250-500","gt-500"]},
    setup_feature_store_db = "recommender_demos",
    setup_feature_store_collection = "set_up_offers"
)
 
# Generate a new options store
ol.generate_options_store(auth,deployment_id)
 
# Update the existing options store
ol.update_options_store(auth,deployment_id)

Parameters Process